How Bacteria in the Mouth and Gut May Affect Bone Health Through Their Tiny Particles

Different bacterial species have distinct pathways to generate EVs, with mechanisms in Gram-negative bacteria being comparatively well characterized. Three major types of vesicles have been described: the classic outer membrane vesicles (OMVs) produced by the bubbling of living cells, outer–inner membrane vesicles (OIMVs) and explosive outer membrane vesicles (EOMVs) produced by cell rupture triggered by endolysins, and OIMVs formed by the transient degradation of PG by autolysins [12,63,64]. The biogenesis mechanism of BEVs in Gram-positive bacteria is not as clear as that in Gram-negative bacteria, and the process of production and release is still controversial [61,63]. Currently, there are three main hypotheses for the mechanism of BEVs secretion through the cell wall of Gram-positive bacteria [65]: (1) vesicle accumulation between the cytoplasmic membrane and the thick PG wall generates pressure until release; (2) enzymatic degradation of PG creates pores that allow vesicle passage; and (3) protein channels or structural cables guide vesicles across the cell wall. Recent studies in archaea have shown that EVs from Methanobrevibacter smithii accumulate in the periplasmic space and are released into the extracellular milieu through localized disruption of the rigid cell wall, a process reminiscent of vesicle passage across Gram-positive walls [66]. Advanced imaging techniques, including cryo-electron tomography (cryo-ET) and super-resolution stochastic optical reconstruction microscopy (STORM), are anticipated to provide deeper mechanistic insights into EV biogenesis in Gram-positive bacteria, where thick PG layers present unique structural challenges.

Vesicle release is not merely a passive byproduct of physical pressure but is also under genetic control as evidenced by studies. In Gram-negative bacteria, for example, outer membrane–PG anchoring proteins such as OmpA and Lpp in Escherichia coli (E. coli) play key stabilizing roles; deletion of these genes significantly increases OMV release [67]. Similarly, in archaea, vesicle production in Haloferax volcanii has been shown to depend on a small Ras-like GTPase (ArvA) and its operon partners, providing direct evidence that prokaryotes can employ conserved membrane-remodeling machineries reminiscent of eukaryotic vesicle trafficking [68]. This suggests that bacteria can actively regulate vesicle output by modulating anchoring proteins. In mammalian cells, EV formation is also governed by genetic factors, with the endosomal sorting complex required for transport (ESCRT) machinery representing a canonical pathway. Together, these examples illustrate that EV biogenesis is subject to tightly controlled genetic regulation.

However, studies on the regulatory mechanisms of Gram-positive bacterial proteins remain limited. Autolysins, which are self-encoded PG hydrolases, have been implicated in vesicle release. For example, S. mutans lytF mutants exhibit altered EV production [69], and in B. subtilis, multiple autolysin genes—including LytC, LytD, LytE, and LytF—have been shown to influence EV secretion [70]. Importantly, distinct biogenetic routes directly influence cargo selectivity—for example, OMVs formed by blebbing exclude cytoplasmic material, whereas OIMVs and EOMVs package both cytoplasmic and membrane-derived cargo (see also Section 3.1.2 and Figure 2). These findings suggest that Gram-positive species may also rely on genetic regulatory networks, in addition to the physical constraints of thick PG walls and pressure-driven release, to control EV biogenesis.

The composition of BEVs is different from that of their parent bacterial cells, with varying degrees of enrichment observed for individual components within BEVs, but their composition differs between Gram-negative and Gram-positive. Vesicles from Gram-negative species typically contain PG fragments, membrane proteins (MPs), lipoproteins, lipopolysaccharides (LPS) or lipooligosaccharides (LOS), and phospholipids, whereas Gram-positive vesicles harbor PGN, MPs, lipoproteins, and lipoteichoic acid (LTA) but lack LPS/LOS.

The membrane protein components enriched in BEVs are influenced by the differential affinity of membranes to various proteins, with outer membrane (OM) integration proteins being preferentially retained. BEVs can also encapsulate diverse compounds, including cytoplasmic proteins, secreted proteins, and nucleic acids (DNA and RNA) [71]. The specific enrichment processes that govern cytoplasmic loading during these distinct vesiculation pathways remain poorly understood and require further investigation [14,72]. Interestingly, many bacteria appear to preferentially release certain components via BEVs rather than retain them within the cell, a phenomenon that may reflect an evolutionarily conserved survival strategy. In Porphyromonas gingivalis, virulence factors such as gingipains and adhesins are selectively enriched in OMVs—often at concentrations significantly higher than in the parent bacterial cell—highlighting EVs as potent mediators of pathogenesis via concentrated toxin delivery [73]. In addition, at different growth stages of bacteria, the physicochemical properties and functionality of BEVs will be different. For example, late-log phase BEVs and stationary phase BEVs of P. gingivalis have more significant pathogenicity in periodontal disease, even almost equal to the pathogenic effect of the parent bacteria [74]. Variations in cargo abundance or structure endow BEVs with unique regulatory roles. For instance, BEV derived from P. gingivalis enrich virulence P. gingivalis proteinase, facilitate their protection against degradation during long-distance delivery, thereby prolonging their activity [75,76]. This kind of BEVs carrying virulence is not just a simple pathogenic effect, it may be related to the host's early establishment of immune tolerance to the microbiota [45].

BEVs are not only diverse in composition, but also selectively package cargo based on multiple factors, reflecting bacteria's ability to adapt to and interact with their environment. This capacity to modulate cargo allows bacteria to enhance survival and adaptability by coordinating or competing across diverse environments. Oral pathogens such as P. gingivalis (gingipains), A. actinomycetemcomitans (leukotoxin), and Fusobacterium nucleatum (F. nucleatum) (FadA) preferentially enrich virulence factors in their vesicles, thereby aggravating host tissue destruction and immune dysregulation. In contrast, commensal species such as Bacteroides employ OMVs to selectively package glycan-degrading enzymes that facilitate nutrient acquisition. For example, proteomic analyses of Bacteroides thetaiotaomicron OMVs demonstrated that glycosidases and proteases are enriched through a negatively charged lipoprotein export signal (LES) motif, enabling these vesicles to act as extracellular "digestive units" that expand polysaccharide utilization. Moreover, the composition of bacterial EVs can vary substantially depending on the strain and culture conditions, reflecting the ability of bacteria to actively tailor EV cargo in response to environmental cues and fine-tune their interactions with the host [77]. The following factors are recognized as major determinants of selective cargo packaging, as depicted in Figure 2: (1) Bacterial growth phase. EV secretion varies across growth stages. In Pseudomonas aeruginosa, EVs derived from biofilms promote biofilm expansion during the exponential phase, while vesicles secreted during stationary or survival phases may reduce biofilm growth, reinforcing their role in cell-to-cell communication [78]. (2) Antibiotics. Sub-minimum inhibitory concentration (sub-MIC) antibiotics can markedly remodel EV cargo. For example, in Geobacter sulfurreducens, ciprofloxacin triggers SOS response and prophage induction, leading to the release of OIMVs enriched in redox-active cytochromes, which enhance Fe(III) reduction capacity, whereas ampicillin primarily induces classical OMVs through outer-membrane blebbing [79]. Thus, sub-MIC antibiotics act as external regulators of vesicle composition and microbial energy metabolism. (3) pH and specific regulators. In Gram-positive Streptococcus mutans, the OpuB transporter modulates vesicle size, lipid composition, and proteomic content. Under neutral pH, the opuB mutant secreted significantly more and smaller vesicles enriched in stress-related proteins such as DNA repair factors, whereas under acidic pH these differences diminished. Together, pH and OpuB regulate MV biogenesis and cargo packaging, enhancing adaptation to cariogenic acidic niches [80]. Furthermore, the growth of Helicobacter pylori under acidic conditions leads to the production of fewer and smaller OMVs, with altered protein enrichment profiles. In contrast, under near-neutral pH, OMVs are enriched in VacA and the putative β-lactamase HcpE, suggesting their potential involvement in pathogenic processes [81]. (4) Nutrient availability. Iron and heme strongly influence OMV cargo. In Neisseria gonorrhoeae, iron limitation alters OMV proteome and physicochemical properties [82]. In P. gingivalis, heme limitation upregulates heme acquisition proteins, whereas heme excess induces a heme efflux system and preferential packaging of moonlighting cytoplasmic proteins into OMVs [83]. These findings underscore nutrient-driven plasticity in EV cargo. (5) Temperature. In Staphylococcus aureus (S. aureus), subtle temperature changes markedly alter EV content. At 40 °C, vesicles are enriched in virulence factors such as α-toxin and αPSMs, enhancing hemolytic activity but reducing RNA abundance and proteome diversity. At 34 °C, vesicles carry more diverse proteins and RNAs and exhibit higher cytotoxicity toward macrophages [84]. The temperature effect may be mediated by changes in membrane fluidity: high temperatures increase fluidity and promote vesicle budding, whereas low temperatures decrease fluidity and reduce vesiculation [84]. (6) Oxidation. In E. coli, OMV cargo selectivity is largely determined by whether proteins are stably tethered to the cell envelope and whether they contain oxidizable residues, with integral MPs being retained in the OM and untethered or oxidized proteins preferentially exported into OMVs under stress [85]. (7) Transcript abundance and RNA selectivity. EV RNA content is shaped by transcript abundance, size, and the localization of encoded proteins. In Lactobacillus johnsonii N6.2, vesicle-associated RNAs were enriched in highly expressed, smaller transcripts encoding membrane-bound or secreted proteins. These RNAs elicited RNA-sensing immune pathways in host cells, emphasizing functional consequences of RNA selectivity [86]. The diverse molecular cargo of BEVs not only shape host–microbe interactions but also mediate complex communication between bacterial species, which will be discussed in the next section.

In addition, EV isolation methods can influence cargo composition, and vesicle size itself is another determinant, typically ranging from 20 to 400 nm [87]. Larger vesicles exhibit greater cargo capacity and distinct loading efficiencies compared with smaller counterparts. Although consensus guidelines such as Minimal Information for Studies of Extracellular Vesicles provide recommendations on isolation strategies and purity standards, this heterogeneity underscores the need to interpret functional studies with caution [88].

The ability of BEVs to selectively package cargo represents a promising strategy, as it allows the reduction in toxic components while enabling the enrichment of immunomodulatory molecules that promote tolerance and host protection. By controlling culture conditions or applying bioengineering strategies, BEVs can be tailored for clinical applications: (i) removing or reducing toxic cargo (e.g., LPS, toxins) to improve safety; (ii) enriching beneficial molecules (e.g., immunoregulatory proteins, metabolites, or therapeutic agents) to enhance efficacy; and (iii) ensuring reproducibility and consistency across production batches. Synthetic biology approaches have been used to engineer E. coli-derived OMVs to carry heterologous antigens or drugs, taking advantage of vesicles' natural capacity to package specific proteins or nucleic acids [89]. Thus, selective cargo packaging, once standardized, may transform BEVs into powerful tools for vaccines, drug delivery, and the treatment of inflammatory and metabolic diseases.

Recent large-scale human microbiome studies revealed that oral microbes can transmit to and colonize the large intestine, establishing the oral cavity as an endogenous reservoir for gut microbial strains [90]. However, this remains observational evidence, and longitudinal follow-up with oral disease stratification is needed. Direct evidence of oral bacteria entering the gut largely comes from animal studies. For example, F. nucleatum introduced to induce periapical periodontitis in mice was detected in the colon within two weeks, accompanied by alterations in GM composition [91]. Given the highly acidic environment of the stomach, direct survival and EV release of oral bacteria along the GIT may be limited. Nevertheless, it cannot be excluded that oral bacteria enter the gut in biofilm aggregates to resist low pH, or under conditions of impaired gastric acid or bile barrier function, where they can colonize and release EVs [92]. Indeed, P. gingivalis and its OMVs have been detected in the GM of patients with IBD. The hydrophobic properties of BEVs protect encapsulated cargo from degradation during long-distance transport and facilitate their uptake by host epithelial and endothelial cells, thereby contributing to inflammation in distal organs such as neuroinflammation [93,94] and bone loss [21,95]. Moreover, the nanoscale size of BEVs allows them to penetrate leaky vasculature, enter the circulation, or cross epithelial and endothelial barriers such as the intestinal epithelium and the BBB.

Crossing the intestinal barrier involves multiple layers: (1) Chemical barrier: BEVs can neutralize AMPs [96] or impair commensal growth [97]; (2) Physical barrier: tight junctions formed by proteins such as occludin and E-cadherin. Pathogenic BEVs (e.g., P. gingivalis OMVs) can downregulate tight-junction proteins and compromise epithelial integrity [98]; immune barrier: epithelial cells recognize BEVs via pattern-recognition receptors such as TLRs (e.g., TLR4) and intracellular receptors such as NOD1, leading to cytokine release [99]. After transcytosis across intestinal epithelial cells, BEVs reach the lamina propria, where they are internalized by resident immune cells, including macrophages and dendritic cells (DCs). Notably, EV uptake by epithelial or immune cells is not merely part of barrier clearance but may also facilitate trans-barrier transport. During cell death processes (apoptosis, necrosis, pyroptosis) or transcytosis, EV cargo can be further released to the opposite side of the barrier [100]. This dual role underscores potential therapeutic applications, such as exploiting commensal-derived EVs to reinforce barrier function by promoting occludin expression or reducing pathogen adhesion.

With respect to the BBB, systematic reviews have suggested that bacterial EVs can indeed cross, though the precise mechanisms remain poorly defined. Most experimental data derive from in vitro permeability models or zebrafish in vivo studies [101,102]. In mammalian EV research, known transport mechanisms include transcytosis [103], receptor-mediated transport [102], and transient disruption of tight junctions. However, current models are largely based on single-layer systems (e.g., transwell, organ-on-chip) and do not fully recapitulate the neurovascular unit. Direct in vivo evidence for gut-derived BEVs crossing the BBB is still lacking [104].

The relationship between bacteria is complex, dynamic, and bidirectional. Bacteria establish and maintain cooperative, symbiotic, and competitive relationships among different genera through quorum sensing (QS), biofilm formation, nutrient acquisition, defense, and other means [95,104,107,108,109]. By packaging signaling molecules, enzymes, and even genetic material, BEVs enable cooperation and competition within microbial communities.

QS serves as a chemical signaling method that orchestrates the behaviors of bacterial communities. It plays a pivotal role in the secretion processes of BEVs from parental bacteria. Proteins related to QS are also released through these BEVs, which are instrumental in the nutrient acquisition processes of bacteria. BEVs are carriers of essential nutrients, including iron ions, vitamins, amino acids, and carbon metabolic components. Furthermore, they transport critical hydrolytic enzymes enabling certain bacteria to access polysaccharides [14,107]. The role of BEVs in environmental stress mitigation involves their content, such as catalase, which degrades reactive oxygen species secreted by host cells. They also contain nucleic acids, including genes for antibiotic resistance; however, the mechanisms of nucleic acid uptake and function in recipient cells remain poorly understood. Additionally, BEVs can function as protective "bait" in hostile environments by binding to bacteriophages and AMPs, thus enhancing the survival of bacterial communities. For instance, BEVs secreted by P. gingivalis can bind to chlorhexidine and extend their protective effect to the biofilm they are in [95].

BEVs also constitute a significant component of the bacterial biofilm matrix, contributing to the biofilm's structural integrity and functional enhancement [109]. For example, P. gingivalis BEVs improve the aggregation and adhesion of diverse oral microbes within dental plaque biofilms [110]. Similarly, BEVs from Pseudomonas carry glycoside hydrolases that can break down various polysaccharides [95]. EVs secreted by Streptococcus mutans may serve as a mechanism of bacterial competition and represent an important dysbiotic factor during the early stages of cariogenic biofilm establishment [111]. In contrast, EV production by P. gingivalis appears to be highly dependent on strain-specific genotypes, and clinically, dental biofilms are likely to harbor multiple EV subtypes derived from diverse P. gingivalis lineages.

Moreover, BEVs contain bioactive molecules that exhibit antifungal and antibacterial properties, which are crucial in bacterial competition and survival. For instance, Bacillus and Staphylococcus release bioactive compounds and lytic enzymes through their BEVs to eliminate competing genera. BEVs from P. gingivalis can suppress the adhesion-related proteins FadA and FomA of F. nucleatum diminishing its ability to adhere to and invade epithelial cells. Additionally, BEVs from Clostridium butyricum significantly reduce the abundance of bacterial pathogens such as E. coli and Shigella flexneri [112].

BEVs engage with human host cells through various mechanisms, including receptor binding, membrane fusion, endocytosis, and phagocytosis. These interactions can profoundly affect cellular functions such as inflammation, apoptosis, necrosis, and autophagy, and also influence the efficiency of cellular uptake of BEVs [113]. The cargo within BEVs can activate a broad spectrum of innate immune responses in the host. This activation promotes cellular immune reactions and cell death via membrane and cytoplasmic pattern recognition receptors (PRRs), such as LPS/LOS or DNA/RNA that activate the TLR pathways, leading to an inflammatory response. Furthermore, the conserved structure of PG can initiate the cytoplasmic NOD pathways, enhancing antibacterial immunity and autophagy [111]. Additional components of BEVs can independently regulate host immunity and promote pathogenesis. For example, internalized endotoxins may induce cellular necrosis. Small interfering RNAs (siRNAs), outer membrane proteins (OMPs), and LPS can modulate host gene expression and immunity, facilitating pathological progression [114,115].

BEV uptake is highly cell-type and receptor dependent, co-determined by vesicle-intrinsic features (size, surface ligands, cargo) and the endocytic repertoire of the recipient. Professional phagocytes (macrophages, DCs) internalize BEVs via phagocytosis and micropinocytosis via various receptors [116]. For example. P. gingivalis OMVs enriched in gingipains engage pattern-recognition receptors TLR2 to enhance uptake and pro-inflammatory signaling in macrophages [117]; Francisella tularensis EVs are taken up by bone-marrow-derived macrophages predominantly via macropinocytosis, as supported by inhibitor studies [118]. These uptake events frequently culminate in downstream cellular outcomes, including inflammation, apoptosis, necrosis, autophagy, or secondary release of vesicular cargo into the extracellular milieu.

Non-phagocytic epithelial cells (e.g., intestinal or oral mucosa) use distinct pathways such as clathrin-mediated endocytosis (CME), caveolin/lipid-raft-dependent endocytosis, and macropinocytosis. For example, Jones observed that intestinal epithelial cells principally acquire Bt OMVs via dynamin-dependent endocytosis followed by intracellular trafficking to LAMP-1 expressing endo-lysosomal vesicles and co-localization with the perinuclear membrane [119].

BEV size further dictates uptake fate: smaller vesicles (<100 nm) typically enter epithelial cells via CME or lipid-rafts and may escape lysosomal degradation to release bioactive cargo into the cytosol; larger vesicles (>200 nm) are more prone to phagocytosis and lysosomal routing [120]. Surface ligands also confer targeting specificity: Bacteroides fragilis OMVs, enriched with polysaccharide A, preferentially target DCs to activate TLR2 and induce regulatory T cell responses [121]; Bacteroides thetaiotaomicron employs a negatively charged LES to sort glycosidases into OMVs, optimizing interaction with gut epithelium and possibly influencing their cellular uptake [84].

Thus, BEV uptake is choreographed by vesicle size, cargo, and ligand receptor compatibility across distinct cell types. From a therapeutic perspective, these mechanisms provide opportunities to optimize engineered BEVs for bone-targeted interventions. To prevent premature degradation, strategies can be designed to bias uptake toward caveolin- or lipid-raft-mediated pathways, which favor cytosolic delivery over lysosomal degradation, thereby enhancing the functional release of therapeutic cargo such as osteogenic miRNAs. Conversely, avoiding excessive uptake by phagocytes (e.g., macrophages) is essential to reduce clearance and preserve bioactivity; this can be achieved by surface modifications that inhibit phagocytosis or by local delivery systems. To improve osteoblast targeting, BEVs can be functionalized with bone-homing ligands (e.g., CXCR4 to exploit SDF-1/CXCL12 gradients, RGD peptides to engage integrins, or bisphosphonate groups to bind mineralized matrix), thereby enriching them in the bone microenvironment and promoting efficient uptake by osteogenic cells. In addition, engineered OMVs can be designed to exploit the Enhanced Permeability and Retention (EPR) effect, thereby facilitating their preferential accumulation at target sites [122].

Oral bacterial OMVs are often enriched in virulence-associated components, and once internalized by epithelial or immune cells, these vesicles can release their cargo, including LPS and other toxic molecules, thereby triggering pro-inflammatory responses.

The fimbrial proteins that make up the fimbriae are components of BEVs, they are crucial for adhesion and possess pro-inflammatory capabilities [151]. These proteins rely on the standard TLR2 pro-inflammatory pathway involving the signal linkers TIRAP and MyD88, as well as a TIRAP/MyD88 independent pro-adhesion pathway mediated by TLR2 [42,152]. Fimbrial proteins are potentially significant in inducing bone destruction and are essential for allowing the parent bacteria to effectively infiltrate monocyte-derived DCs. However, it remains unclear whether the fimbrial proteins in BEVs assist in cell recognition by bacteria. Fimbrial proteins can prompt fibroblasts, epithelial cells, and monocyte/macrophages to produce inflammatory cytokines such as IL-1α, IL-1β, IL-6, and TNF-α. IL-1 and TNF-α are crucial local regulators in bone remodeling and play significant roles as immune mediators in periodontal inflammation and alveolar bone destruction [153].

Gingipain is a family of cysteine proteases, also known as gingipains, including arginine-specific gingipains (Rgp, including RgpA and RgpB) and lysine-specific gingipains (Kpg). In different strains of P. gingivalis, mature gingipains can be transported and docked on the bacterial OM or released into the extracellular environment, and have been confirmed to be enriched in P. gingivalis EVs [154]. Gingipains are responsible for 85% of the extracellular protein hydrolysis activity of P. gingivalis, by participating in the activation of host matrix metalloproteinases, stimulating oral epithelial cells to produce IL-6 and gingival fibroblasts to produce IL-8, and promoting the recruitment of polymorphonuclear neutrophils (PMN) through complement system activation, and also by cutting several TCRs and protein hydrolysis inactivation factors such as IFN-γ, IL-4, IL-5, and IL-12 to hinder host immunity, causing dysregulated immune response and inflammation [42]. Gingipains can not only regulate immune abnormalities that can cause bone loss [155], but can also directly act on osteoblasts and osteoclasts. Gingipains mediate osteoblast apoptosis in vitro experiments, possibly through two pathways: one is to inhibit the expression of integrin β1 in HCOs cell and MC3T3-E1 cell, accompanied by a decrease in active RhoA, mediating F-actin rupture and cell apoptosis [156]; another is that gingipains induce BH3 interacting domain death agonist and P53 expression upregulation by inhibiting the extracellular regulated protein kinases (ERK) and protein kinase B (PKB) pathway of MC3T3-E1 cell, causing apoptosis [157]. Although, gingipains cannot directly activate osteoclasts [158], but can upregulate osteoclast-related genes and proteins in a time-dependent manner, including cathepsin K (Ctsk), matrix metalloproteinase 9 (Mmp 9), nuclear factor of activated T-cells 1 (Nfatc 1) and tartrate-resistant acid phosphatase 5 (Acp 5), by promoting the expression of integrin αvβ, enhancing the bone resorption function of osteoclasts induced by RANKL [159], and at the same time can hydrolyze the bone protective substance secreted by osteoblasts in the extracellular [160,161], change the RANKL/OPG ratio, enhance the differentiation of osteoclasts under the induction conditions of IL-1β and TNF-α.

T. denticola and T. forsythia are important members of the red complex of periodontal disease, but unfortunately, their pathogenic mechanisms causing bone loss in periodontal disease are still poorly understood [162,163]. It is known that T. denticola harms include the destruction of the host cell extracellular matrix, tissue infiltration, and destruction of the host cell membrane [164,165], accompanied by the dysregulation of host immune regulatory factors [166]. T. forsythia produces inflammatory signals through TLR2, leading to alveolar bone loss related to Th2 cells, among which, BspA protein not only plays a role in adhesion, but also triggers the release of pro-inflammatory cytokines from monocytes and chemokines from osteoblasts, leading to inflammation and bone resorption [167]. However, the pathogenicity of T. forsythia depends on the role of other bacteria, and bone loss cannot be observed in GF mice. Interestingly, although the BEVs of both induce Th cells polarization in bone-marrow-derived dendritic cells (BMDCs), the mechanism is complex, T. denticola EVs mediate Th17 polarization, while T. forsythia EVs mediate Th1 polarization, which is inconsistent with the result of T. forsythia GroEL cooperating with IL-17 to induce bone loss, which may be related to different BEVs enrichment components [168]. Proteases (such as dentin) on T. denticola EVs widely hydrolyze cells to produce IL-1β, IL-6, IL-8, and TNF-α. The S-layer glycan structure on T. forsythia EVs protects parent bacteria from being recognized by BMDCs [169]. Therefore, the mechanism of immune dysregulation caused by these BEVs may be more important in bone loss.

F. nucleatum is also a rich periodontal bacterium, present in the oral cavity of both periodontally healthy and diseased patients [170]. F. nucleatum EVs contain a variety of antigenic virulence factors, including FadA, MORN 2, and YadA-like domains [171]. The virulence factor FadA allows F. nucleatum to survive under stress conditions, form biofilms, accumulate in diseased periodontal sites, bind to host cells, and induce periodontal bone loss. However, amyloid-like FadA does not seem to independently cause the manifestation of OP/bone loss, but enhances the virulence potential of F. nucleatum, because F. nucleatum lacking FadA did not detect significant periodontal bone loss [170]. Although, FadA has the function of regulating cell Wnt/β-catenin [172], but no abnormal function of bone lineage cells has been found yet. But some researchers have found that FadA plays an important role in the pathogenesis of RA, which will be discussed later.

While research has well-established the direct relationship between bacteria and alveolar bone resorption in the oral pathological microenvironment, the association of BEVs with systemic bone loss requires more direct evidence. Six groups of bacteria species are highly associated with biofilm formation and maturation, contributing to periodontal bone loss. Among these, the 'Red complex,' which forms at a later stage, is associated with the progression of periodontitis. This complex comprises P. gingivalis, T. denticola [165,173,174], and T. forsythia [175,176], and could release EVs into oral environment. Under the dual stimulation of dysregulated microbiota and excessive immune response, the homeostasis of alveolar bone tissue is disrupted, leading to bone resorption and progression of periodontitis. Due to their nanoparticle size, adhesiveness, and proteolytic properties, BEVs can migrate through host tissues, disrupt tight epithelial connections, and deliver bacterial virulence factors to immune cells in the underlying tissues.

BEVs derived from Filifactor alocis (F. alocis) inhibit the differentiation and mineralization of BMSCs osteoblasts in a dose-dependent manner, leading to downregulation of the alkaline phosphatase, osteocalcin, and type I collagen [23]. BEVs modulate the MAPK NF-κB signaling pathway via TLR2, which leads to reduced expression of osteogenic marker genes during the osteogenic differentiation process of BMSCs, thereby inhibiting osteogenic differentiation [177,178,179].

In addition to inhibiting osteogenic processes, BEVs also facilitate osteoclastic bone resorption by promoting osteoclastic activation and differentiation, inducing macrophage polarization, and facilitating the release of inflammatory mediators, as depicted in Figure 3. (1) F. alocis EVs predominantly activate the TLR2 pathway, leading to the phosphorylation of NF-κB, ERK, JNK, and p38, which in turn stimulates the expression of osteoclastogenic transcription factors NFATc1 and c-Fos. While other TLRs like TLR4, TLR9, or the intracellular receptor NOD has also been reported [23,180], TLR2 appears to be the primary immune receptor for BEVs on the cell surface. The effectiveness of BEVs on osteoclast differentiation is notably reduced in TLR2-deficient osteoclast precursors [25], which is associated with abundance of LPS and lipoproteins in BEVs. (2) F. nucleatum EVs effectively recruit macrophages to the alveolar bone, promoting their differentiation into a pro-inflammatory M1 phenotype in a time- and dose-dependent manner. This increases the release of IL-1β, IL-6, and TNF-α, driving bone resorption during the early stages of disease [181,182]. The polarizing effect of F. nucleatum EVs on macrophages remains a subject of debate, with some studies suggesting a potential role in M2 polarization, indicating that further verification is needed [183]. (3) BEVs stimulate macrophages, DCs, epithelial cells, endothelial cells, and others to secrete pro-inflammatory factors such as TNF-α, IL-1, and IL-6. Note that the immune responses were dependent on EVs concentration. At low concentrations, these EVs provoke the secretion of TNFα, IL-1β, and IL-8 in unprimed cells, recruit inflammatory cells in vivo, and trigger macrophages to produce a heightened immune response upon a second exposure. Conversely, high concentrations of P. gingivalis EVs induce less inflammation initially due to LPS tolerance and inflammatory unresponsiveness, but lead to strong inflammasome activation upon subsequent exposure. This flexible immune stimulation might facilitate manipulation and disruption of host immune responses, potentially accelerating disease progression alongside pathogens such as T. denticola and T. escherichia [154,184].